In a hierarchical model, is it a problem if my predictors have different ranges for different participants?
My data is as follows: I have ~ 400 participants who did several (~80 per person) behavioural measurements and also reported their mood. I want to make a model like this:
behaviour ~ mood + vars_of_no_interest + (mood + vars_of_no_interest|subject)
Now what I’m wondering whether it’s problematic or not: because I had no experimental control over the range of mood measurements participants make (e.g. some might rate their happiness from 0 to 50, while others might rate it from 50 to 10 or yet others from 90 to 100), this regressor does not have the same range across participants.
Is this a problem?
I’d be very grateful for any pointers.
I’m not quite sure I understand - so the variable
mood in your model consists of a variety of different scales that were given to many participants? In other words, some would get one scale where the score had a min-max range of 0 - 50, and others would get another scale where the min-max range was 90-100? Or, were they all given the same scale and simply picked out different scores?
Sorry, I wasn’t quite clear. Everyone was given the same scale from 0 to 100 and they rated their mood repeatedly (with a slider underneath a question e.g. ‘how happy are you right now?’). And it just happened that say some people were more happy than others (scoring say between 50 and 100 vs a more sad person scoring between 0 and 50).
I don’t see any problem here. Seems like a good place to use a hierarchical model like you set up with varying slopes for